Skill Premium Divergence: The Roles of Trade, Capital and Demographics Sang-Wook (Stanley) Cho Juli´ an P. D´ ıaz School of Economics Department of Economics Australian School of Business Quinlan School of Business University of New South Wales Loyola University Chicago Prepared for the Bank of Estonia October 2017
42
Embed
Skill Premium Divergence: The Roles of Trade, Capital and Demographics
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Skill Premium Divergence: The Roles of Trade, Capital andDemographics
Sang-Wook (Stanley) Cho Julian P. DıazSchool of Economics Department of Economics
Australian School of Business Quinlan School of BusinessUniversity of New South Wales Loyola University Chicago
Prepared for the Bank of EstoniaOctober 2017
dslfkajsjk hjgjasgdasgdhasghjasg
What drives the patterns of the skill premium?
– No unanimous consensus reached on which factors definitively drivethe patterns of the skill premium (ws/wu), but a few hypotheseshave emerged as prime candidates:
– Technological change that favors skilled workers that arecomplementary to capital than unskilled workers.
– The expansion of trade that encourages production in sectorsthat use a particular type of labor intensively.
– Changes in the skill composition of the population that makeskilled/unskilled labor more abundant/scarce.
– Problem: most articles focus on one factor at a time, neglectingtheir potential interactions.
– Aimed at accounting episodes of skill premium increases ordecreases.
dslfkajsjk hjgjasgdasgdhasghjasg
What drives the patterns of the skill premium?
– No unanimous consensus reached on which factors definitively drivethe patterns of the skill premium (ws/wu), but a few hypotheseshave emerged as prime candidates:
– Technological change that favors skilled workers that arecomplementary to capital than unskilled workers.
– The expansion of trade that encourages production in sectorsthat use a particular type of labor intensively.
– Changes in the skill composition of the population that makeskilled/unskilled labor more abundant/scarce.
– Problem: most articles focus on one factor at a time, neglectingtheir potential interactions.
– Aimed at accounting episodes of skill premium increases ordecreases.
dslfkajsjk hjgjasgdasgdhasghjasg
What drives the patterns of the skill premium?
– No unanimous consensus reached on which factors definitively drivethe patterns of the skill premium (ws/wu), but a few hypotheseshave emerged as prime candidates:
– Technological change that favors skilled workers that arecomplementary to capital than unskilled workers.
– The expansion of trade that encourages production in sectorsthat use a particular type of labor intensively.
– Changes in the skill composition of the population that makeskilled/unskilled labor more abundant/scarce.
– Problem: most articles focus on one factor at a time, neglectingtheir potential interactions.
– Aimed at accounting episodes of skill premium increases ordecreases.
dslfkajsjk hjgjasgdasgdhasghjasg
What drives the patterns of the skill premium?
– No unanimous consensus reached on which factors definitively drivethe patterns of the skill premium (ws/wu), but a few hypotheseshave emerged as prime candidates:
– Technological change that favors skilled workers that arecomplementary to capital than unskilled workers.
– The expansion of trade that encourages production in sectorsthat use a particular type of labor intensively.
– Changes in the skill composition of the population that makeskilled/unskilled labor more abundant/scarce.
– Problem: most articles focus on one factor at a time, neglectingtheir potential interactions.
– Aimed at accounting episodes of skill premium increases ordecreases.
dslfkajsjk hjgjasgdasgdhasghjasg
What Do We Do?
– We construct a static general equilibrium model that incorporatesall three factors simultaneously.
– Our model allows us to assess the individual and joint effects ofeach factor on the skill premium.
– Since our framework includes factors that affect the demand andsupply for each type of labor, their overall effect could theoreticallylead to skill premium increases or decreases.
– To assess whether our model’s predictions are in line with the data,we apply it to account for the patterns of the skill premium in thethree Baltic states: Estonia, Latvia and Lithuania.
dslfkajsjk hjgjasgdasgdhasghjasg
Why the Baltics?
– As they transitioned from centrally-planned to free-market systems:
– Aggressively opened their economies to the rest of the world.
– Experienced rapid expansions in their stocks of capital.
– Went through significant changes in the skill composition oftheir labor forces.
– Moreover, they displayed very similar levels of the skill premium in1995 (around 1.8).
– Since then, the Baltic skill premia took divergent patterns.
“Skill” definition is educational attainment: skilled workers are those with tertiaryeducation, unskilled workers are those with non-tertiary education.
Estonia Latvia Lithuania Estonia Latvia Lithuania14.5% -10.9% 29.9% 34.0% 8.3% 41.8%
Source: AMECO database
dslfkajsjk hjgjasgdasgdhasghjasg
– Summarizing:
– Equipment capital grew at the fastest rate in Latvia and theslowest in Estonia.
– Skilled population grew faster in Lithuania and Estonia and moreslowly in Latvia.
– Unskilled population shrank faster in Estonia and Lithuania andmore slowly in Latvia.
– Terms of trade in the goods sector improved in Estonia andLithuania, but worsened in Latvia.
dslfkajsjk hjgjasgdasgdhasghjasg
The Model
– Main ingredients:
– Two-country static general equilibrium model: Baltic state andROW.
– Impose small-economy assumption on Baltic state.
– Agents: households, firms, government, rest of the world.
– Two types of labor: skilled (`s) and unskilled (`u).Two types of capital: equipment (ke) and structures (kz).
– Multi-sector model: goods and services with different skillintensities.
dslfkajsjk hjgjasgdasgdhasghjasg
Flow chart aggregate economy
Households Producers
Government
Rest of theWorld
direct taxes
capital (equipment and structures)labor (skilled and unskilled)
goods and services
indirect taxestariffs
goods and services
inter/intraindustry
transactions
goods and services
importedcomponentssavings
(trade balance)
dslfkajsjk hjgjasgdasgdhasghjasg
Production plans
domesticproduction
firms
finalproduction
firms
intermediateinputs
labor(skilled and unskilled)
importedgoods(m)
capital (equipment and structure)
dslfkajsjk hjgjasgdasgdhasghjasg
Final goods production
– Final production in sector i — Armington aggregator:
yi = φi
[δiy
ρm,id ,i + (1− δi )y
ρm,if ,i
] 1ρm,i
yf ,i is the imported component with foreign price pf ,i , which theBaltic state takes as given.
– ρm,i is the parameter that governs the elasticity of substitutionbetween domestic and imported components in sector i
– Imports of commodity i are subject to ad-valorem tariff rate τf ,i(negligible for the Baltics)
dslfkajsjk hjgjasgdasgdhasghjasg
Final goods production
– Final production in sector i — Armington aggregator:
yi = φi
[δiy
ρm,id ,i + (1− δi )y
ρm,if ,i
] 1ρm,i
yf ,i is the imported component with foreign price pf ,i , which theBaltic state takes as given.
– ρm,i is the parameter that governs the elasticity of substitutionbetween domestic and imported components in sector i
– Imports of commodity i are subject to ad-valorem tariff rate τf ,i(negligible for the Baltics)
dslfkajsjk hjgjasgdasgdhasghjasg
Domestic goods production
– Domestic component production in sector i — Nested CES:
yd,i = min
x1,ia1,i
, . . . ,xn,ian,i
, βikαi
z,i
[λi
[µik
ρe,i + (1− µi )`
ρs,i
]σρ
+ (1− λi )`σu,i] 1−αi
σ
– αi , µi and λi are the share parameters of inputs in value added
– ρ captures the elasticity of substitution between equipment andskilled labor; and σ governs the elasticity of substitution betweenunskilled labor and equipment or skilled labor
– “Capital-skill complementarity” when σ > ρ (or 11−σ >
11−ρ)
dslfkajsjk hjgjasgdasgdhasghjasg
Household
– Utility function of household of type j :
Uj =
ζj(∑
i∈Iθji c
ηi ,j + θjinv (cinv ,j + cb,j)
η
)ψη
+ (1− ζj)(Lj − `j)ψ
1ψ
subject to
∑i∈I
pici ,j + pinv (cinv ,j + cb,j) = (1− t jd)(wj`j + re ke,j + rz kz,j)
– cinv ,j and cb,j refer to domestic savings and government-issuedbonds.
dslfkajsjk hjgjasgdasgdhasghjasg
Skill Premium in the Model
– Using the FOCs, we can derive the expression for the skill premium:
ws
wu=
λi (1− µi )
1− λi
[µi
(ke,i`s,i
)ρ+ (1− µi )
]σ−ρρ
(`s,i`u,i
)σ−1
– Expressing in growth rates terms (γ):
γSP ' µi (σ − ρ)
(ke,i`s,i
)ρ(γke,i − γ`s,i ) + (σ − 1)(γ`s,i − γ`u,i )
– First term: if σ > ρ, increases in equipment capital lead toincreases in the skill premium in presence of equipment-skillcomplementarity.
– Second term: relative growth rates of skilled and unskilled laboraffect the skill premium.
– Cross-sector reallocation of factors—due to trade shocks, forexample—will also affect skill premium.
dslfkajsjk hjgjasgdasgdhasghjasg
Skill Premium in the Model
– Using the FOCs, we can derive the expression for the skill premium:
ws
wu=
λi (1− µi )
1− λi
[µi
(ke,i`s,i
)ρ+ (1− µi )
]σ−ρρ
(`s,i`u,i
)σ−1
– Expressing in growth rates terms (γ):
γSP ' µi (σ − ρ)
(ke,i`s,i
)ρ(γke,i − γ`s,i ) + (σ − 1)(γ`s,i − γ`u,i )
– First term: if σ > ρ, increases in equipment capital lead toincreases in the skill premium in presence of equipment-skillcomplementarity.
– Second term: relative growth rates of skilled and unskilled laboraffect the skill premium.
– Cross-sector reallocation of factors—due to trade shocks, forexample—will also affect skill premium.
dslfkajsjk hjgjasgdasgdhasghjasg
Skill Premium in the Model
– Using the FOCs, we can derive the expression for the skill premium:
ws
wu=
λi (1− µi )
1− λi
[µi
(ke,i`s,i
)ρ+ (1− µi )
]σ−ρρ
(`s,i`u,i
)σ−1
– Expressing in growth rates terms (γ):
γSP ' µi (σ − ρ)
(ke,i`s,i
)ρ(γke,i − γ`s,i ) + (σ − 1)(γ`s,i − γ`u,i )
– First term: if σ > ρ, increases in equipment capital lead toincreases in the skill premium in presence of equipment-skillcomplementarity.
– Second term: relative growth rates of skilled and unskilled laboraffect the skill premium.
– Cross-sector reallocation of factors—due to trade shocks, forexample—will also affect skill premium.
dslfkajsjk hjgjasgdasgdhasghjasg
Skill Premium in the Model
– Using the FOCs, we can derive the expression for the skill premium:
ws
wu=
λi (1− µi )
1− λi
[µi
(ke,i`s,i
)ρ+ (1− µi )
]σ−ρρ
(`s,i`u,i
)σ−1
– Expressing in growth rates terms (γ):
γSP ' µi (σ − ρ)
(ke,i`s,i
)ρ(γke,i − γ`s,i ) + (σ − 1)(γ`s,i − γ`u,i )
– First term: if σ > ρ, increases in equipment capital lead toincreases in the skill premium in presence of equipment-skillcomplementarity.
– Second term: relative growth rates of skilled and unskilled laboraffect the skill premium.
– Cross-sector reallocation of factors—due to trade shocks, forexample—will also affect skill premium.
dslfkajsjk hjgjasgdasgdhasghjasg
Skill Premium in the Model
– Using the FOCs, we can derive the expression for the skill premium:
ws
wu=
λi (1− µi )
1− λi
[µi
(ke,i`s,i
)ρ+ (1− µi )
]σ−ρρ
(`s,i`u,i
)σ−1
– Expressing in growth rates terms (γ):
γSP ' µi (σ − ρ)
(ke,i`s,i
)ρ(γke,i − γ`s,i ) + (σ − 1)(γ`s,i − γ`u,i )
– First term: if σ > ρ, increases in equipment capital lead toincreases in the skill premium in presence of equipment-skillcomplementarity.
– Second term: relative growth rates of skilled and unskilled laboraffect the skill premium.
– Cross-sector reallocation of factors—due to trade shocks, forexample—will also affect skill premium.
dslfkajsjk hjgjasgdasgdhasghjasg
Calibration and Data
– Our model is calibrated so that it matches Estonian, Latvian andLithuanian data in 1995.
– Most of the parameters can be directly calibrated from SocialAccounting Matrices we construct for each country using theoptimality and market-clearing conditions. SAM
– Elasticity parameters taken from the literature. We later runsensitivity checks on those values.
Parameter Value Corresponding Elasticity Implied Elasticity
ρm,i 0.827 Import elasticity of substitution 5.78ρx 0.9 Export elast. of subst. 10ρ -0.5 Equipment-skilled labor elast. 0.67σ 0.4 Equipment-unskilled labor elast. 1.67η -1 Consumption goods elast. of subst. 0.5ψ -0.25 Consumption-leisure elast. of subst. 0.8
dslfkajsjk hjgjasgdasgdhasghjasg
Numerical Experiments: Description
– Comparative statics experiments:
– ToT experiment: replicates changes in ToT experienced byBaltic states between 1995 and 2008 by varying foreign prices.
– Capital-deepening experiment: replicates increases in both typesof capital stocks by increasing their endowments.
– Skill supply experiment: increase endowments of skilled andunskilled labor to match those observed in the Baltics.
– Joint experiment: simulates all three shocks simultaneously.
– In each instance, we compare original and new equilibria and assessthe effects of each shock on the skill premium.
dslfkajsjk hjgjasgdasgdhasghjasg
Results - SP changes
Estonia Latvia Lithuania
Data -20.2 16.3 -13.1Experiment
ToT -2.2 -1.9 -1.3Capital deepeningSkill composition
– ToT experiment predicts skill premium declines due to H-O type reallocation ofresources towards unskilled-intensive sectors where Baltics enjoy comparativeadvantages. ToT
– Capital deepening predicts skill premium increases due to equipment-skillcomplementarity. Note that the effect is the strongest in Latvia, which recordedthe largest increase in equipment capital.
– Skill composition predicts skill premium decreases due to skilled labor becomingmore abundant and unskilled labor scarcer.
– Joint experiment: skill premium decreases in Estonia and Lithuania, andincrease in Latvia.
– Overall, our model is able to replicate Baltic divergence. Qualitative results arein line with data. Moreover, quantitative predictions fairly accurate.
– ToT experiment predicts skill premium declines due to H-O type reallocation ofresources towards unskilled-intensive sectors where Baltics enjoy comparativeadvantages.
– Capital deepening predicts skill premium increases due to equipment-skillcomplementarity. Note that the effect is the strongest in Latvia, which recordedthe largest increase in equipment capital. capital
– Skill composition predicts skill premium decreases due to skilled labor becomingmore abundant and unskilled labor scarcer.
– Joint experiment: skill premium decreases in Estonia and Lithuania, andincrease in Latvia.
– Overall, our model is able to replicate Baltic divergence. Qualitative results arein line with data. Moreover, quantitative predictions fairly accurate.
– ToT experiment predicts skill premium declines due to H-O type reallocation ofresources towards unskilled-intensive sectors where Baltics enjoy comparativeadvantages.
– Capital deepening predicts skill premium increases due to equipment-skillcomplementarity. Note that the effect is the strongest in Latvia, which recordedthe largest increase in equipment capital.
– Skill composition predicts skill premium decreases due to skilled labor becomingmore abundant and unskilled labor scarcer. skill
– Joint experiment: skill premium decreases in Estonia and Lithuania, andincrease in Latvia.
– Overall, our model is able to replicate Baltic divergence. Qualitative results arein line with data. Moreover, quantitative predictions fairly accurate.
dslfkajsjk hjgjasgdasgdhasghjasg
Results - SP changes
Estonia Latvia Lithuania
Data -20.2 16.3 -13.1Joint experiment -22.4 9.8 -35.3
– ToT experiment predicts skill premium declines due to H-O type reallocation ofresources towards unskilled-intensive sectors where Baltics enjoy comparativeadvantages.
– Capital deepening predicts skill premium increases due to equipment-skillcomplementarity. Note that the effect is the strongest in Latvia, which recordedthe largest increase in equipment capital.
– Skill composition predicts skill premium decreases due to skilled labor becomingmore abundant and unskilled labor scarcer.
– Joint experiment: skill premium decreases in Estonia and Lithuania, andincrease in Latvia.
– Overall, our model is able to replicate Baltic divergence. Qualitative results arein line with data. Moreover, quantitative predictions fairly accurate.
dslfkajsjk hjgjasgdasgdhasghjasg
Results - SP changes
Estonia Latvia Lithuania
Data -20.2 16.3 -13.1Joint experiment -22.4 9.8 -35.3
– ToT experiment predicts skill premium declines due to H-O type reallocation ofresources towards unskilled-intensive sectors where Baltics enjoy comparativeadvantages.
– Capital deepening predicts skill premium increases due to equipment-skillcomplementarity. Note that the effect is the strongest in Latvia, which recordedthe largest increase in equipment capital.
– Skill composition predicts skill premium decreases due to skilled labor becomingmore abundant and unskilled labor scarcer.
– Joint experiment: skill premium decreases in Estonia and Lithuania, andincrease in Latvia.
– Overall, our model is able to replicate Baltic divergence. Qualitative results arein line with data. Moreover, quantitative predictions fairly accurate.
dslfkajsjk hjgjasgdasgdhasghjasg
Sensitivity Analysis 1 – Model Performance in Shorter Horizons
1995-2000 2000-2008
SP changes Estonia Latvia Lithuania Estonia Latvia Lithuania
– We want to see if our model can also account for the skill premium patternswithin the time span as their evolution were non-monotonic (initial rise andsubsequent fall).
– Our model reproduces an inverse U-shaped evolution of the skill premium inLatvia and Lithuania, but not in Estonia.
– In fact, the labor supply factors seem to have a lagged effect on the skillpremium in Estonia, which our model cannot capture.
– Country-specific institutional factors in the labor market – Toomet (2011)
dslfkajsjk hjgjasgdasgdhasghjasg
Sensitivity Analysis 1 – Model Performance in Shorter Horizons
1995-2000 2000-2008
Experiment Estonia Latvia Lithuania Estonia Latvia Lithuania
– We want to see if our model can also account for the skill premium patternswithin the time span as their evolution were non-monotonic (initial rise andsubsequent fall).
– Our model reproduces an inverse U-shaped evolution of the skill premium inLatvia and Lithuania, but not in Estonia.
– In fact, the labor supply factors seem to have a lagged effect on the skillpremium in Estonia, which our model cannot capture.
– Country-specific institutional factors in the labor market – Toomet (2011)
dslfkajsjk hjgjasgdasgdhasghjasg
Sensitivity Analysis 2 – Changes in Technology ParametersChange in Skill Premium
– With the Cobb-Douglas specification, we abstract from capital-skillcomplementarity and the skill premium only depends on the relative skill supplyand trade channels.
– Our qualitative results crucially hinge on the capital-skill complementaritychannel.
dslfkajsjk hjgjasgdasgdhasghjasg
Sensitivity Analysis 3 – Changes in Preference ParametersChange in Skill Premium
– The lower the elasticity between consumption and leisure, the larger thechanges in the skill premium.
– Labor-leisure margin changes the relative magnitude of the skill premiumchanges. Our qualititavie results are still maintained under an inelastic laborsupply assumption.
dslfkajsjk hjgjasgdasgdhasghjasg
Conclusions
– We construct a static general equilibrium model to account for theevolution of the skill premium.
– Our model incorporates forces that have a biased effect on thedemand for skilled and unskilled labor: international trade andcapital-skill complementarity.
– Unlike the large majority of recent articles, our model alsoincorporate factors that affect the supply of labor.
– To assess the quantitative validity of the model’s predictions, weapply it to account for the divergent patterns of the skill premiumin the Baltic states between 1995 and 2008.
dslfkajsjk hjgjasgdasgdhasghjasg
Conclusions
– Incorporating all forces together, we can qualitatively account forthe Baltic skill premium divergence with a fair degree ofquantitative success.
– For shorter horizons, our model is able to replicate the initial riseand subsequent fall of the skill premium in Latvia and Lithuania,although not for Estonia.
– Our results are robust to the choices of trade and preferenceelasticities, while the sensitivity analyses highlight the capital-skillcomplementarity channel.
– ToT shocks lead to increases in trade in Estonia and Lithuania in both sectors.As a result, final output rises and wages go up. Sectoral reallocation occursfrom skilled (service) sector towards unskilled (goods) sector.
– In Latvia, overall trade falls, leading to lower final output and lower wages. Inthe goods sector, domestic production increases to substitute higher pricedimports.
– Despite mixed effects in the hours worked by different skill types, total hoursworked rise for all countries. This leads to higher aggregate output – final anddomestic.
– As unskilled labor become scarcer, its wage rises. Vice versa for skilled labor.
– Changes in the relative wages translate into changes in the relative demand forskilled and unskilled workers. back